Folks, let’s not turn this thread into a personal disputes arena… Please?
Control Resonant crash with HDR+Wayland on the initial settings screen still happens with 620 (first reported in 615):
i run the game with no launch options at all and i played for 14 hours the other day zero freeze zero issues just this map
I can’t edit it anymore but I forgot that NVML reports a GPU only capable of 8x PCIe max width as having a 16x max width.
Just flag his posts as spam because that’s what they essentially are. Useless clutter in every single thread.
can you please provide any other information as this is entirely unhelpful when others and Nvidia can easily replicate the issue but your system magically doesnt
as i said previously
or proton version that might avoid the issue?
literally any info at all besides “works for me”
I’m getting kernel errors when disabling display port screens
[Mon Nov 9 00:12:40 2026] nv_flip == NULL
[Mon Nov 9 00:12:40 2026] WARNING: nvidia-drm/nvidia-drm-crtc.h:371 at __nv_drm_handle_flip_event+0x1f8/0x210 [nvidia_drm], CPU#20: nvidia-modeset//927
[Mon Nov 9 00:12:40 2026] Modules linked in: wireguard libcurve25519 ip6_udp_tunnel udp_tunnel snd_usb_audio snd_usbmidi_lib snd_ump snd_rawmidi mc snd_ctl_led nf_conntrack_netlink xt_nat veth xt_conntrack bridge stp llc xfrm_user xfrm_algo xt_set ip_set xt_addrtype overlay xt_MASQUERADE rfcomm snd_seq_dummy snd_hrtimer snd_seq snd_seq_device nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 xt_cgroup xt_mark xt_owner xt_tcpudp ipt_REJECT nf_reject_ipv4 xt_multiport nft_compat x_tables nf_tables algif_hash algif_skcipher af_alg bnep nct6775 nct6775_core hwmon_vid nvidia_drm(OE) snd_hda_codec_alc882 snd_hda_codec_nvhdmi iwlmvm snd_hda_codec_realtek_lib snd_hda_codec_generic nvidia_modeset(OE) snd_hda_codec_hdmi snd_hda_intel mac80211 snd_hda_codec snd_hda_core ptp pps_core snd_intel_dspcfg raid1 libarc4 btusb snd_intel_sdw_acpi btmtk snd_hwdep drm_ttm_helper r8169 ttm nvidia_uvm(OE) btrtl snd_pcm ee1004 realtek iwlwifi btbcm drm_display_helper phy_package btintel sp5100_tco snd_timer
[Mon Nov 9 00:12:40 2026] intel_rapl_msr vboxnetadp(OE) vfat mdio_devres vboxnetflt(OE) fat intel_rapl_common cfg80211 bluetooth libphy cec snd i2c_piix4 mousedev joydev md_mod rapl wmi_bmof pcspkr i2c_smbus video soundcore zenpower(OE) rfkill mdio_bus mac_hid vboxdrv(OE) kvm_amd ccp nvidia(OE) gcm kvm aesni_intel gf128mul dm_mod irqbypass sg aead i2c_dev ntsync crypto_user hid_logitech_hidpp nvme nvme_core mxm_wmi nvme_keyring nvme_auth wmi hid_logitech_dj
[Mon Nov 9 00:12:40 2026] CPU: 20 UID: 0 PID: 927 Comm: nvidia-modeset/ Tainted: G W OE 7.2.14-zen1-1-zen #1 PREEMPT(full) d24bc5148e586fc1a6a8333455ab6f14397ed471
[Mon Nov 9 00:12:40 2026] Tainted: [W]=WARN, [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
[Mon Nov 9 00:12:40 2026] Hardware name: Micro-Star International Co., Ltd. MS-7C35/MEG X570 UNIFY (MS-7C35), BIOS A.I0 10/24/2023
[Mon Nov 9 00:12:40 2026] RIP: 0010:__nv_drm_handle_flip_event+0x1f8/0x210 [nvidia_drm]
[Mon Nov 9 00:12:40 2026] Code: 80 3d 33 b0 57 ff 00 74 92 48 89 ef e8 c1 9b 4a d3 eb 88 4c 8b 6c 24 30 4c 8b 7c 24 40 e9 98 fe ff ff 4c 89 e7 e8 08 8e b0 d3 <0f> 0b e9 81 fe ff ff 48 8d bd b0 00 00 00 e8 95 9b 4a d3 e9 27 ff
[Mon Nov 9 00:12:40 2026] RSP: 0018:ffffd14202087dd8 EFLAGS: 00010282
[Mon Nov 9 00:12:40 2026] RAX: ffff88e1f428c818 RBX: ffff88e1f428c818 RCX: ffff88e1d3bd0ac0
[Mon Nov 9 00:12:40 2026] RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff88e1f428c840
[Mon Nov 9 00:12:40 2026] RBP: ffff88e1f428c800 R08: 0000000000000000 R09: ffffffffc2c93b9c
[Mon Nov 9 00:12:40 2026] R10: fffff88e1486d700 R11: ffff88e1c0042700 R12: ffff88e1f428c840
[Mon Nov 9 00:12:40 2026] R13: ffff88e1e1390000 R14: ffff88e1d3bd0800 R15: ffffffffc0b6b5c0
[Mon Nov 9 00:12:40 2026] FS: 0000000000000000(0000) GS:ffff88ed169dd000(0000) knlGS:0000000000000000
[Mon Nov 9 00:12:40 2026] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[Mon Nov 9 00:12:40 2026] CR2: 000026a597470000 CR3: 000000089f422000 CR4: 0000000000350ef0
[Mon Nov 9 00:12:40 2026] Call Trace:
[Mon Nov 9 00:12:40 2026]
[Mon Nov 9 00:12:40 2026] ? srso_return_thunk+0x5/0x5f
[Mon Nov 9 00:12:40 2026] ? __pfx__main_loop+0x10/0x10 [nvidia_modeset 3fd61761179da386f97f306a38b0e56e9d555b16]
[Mon Nov 9 00:12:40 2026] ? srso_return_thunk+0x5/0x5f
[Mon Nov 9 00:12:40 2026] nv_drm_event_callback+0xaa/0xb0 [nvidia_drm 2e615ec2f99f79ae71a6a4d1e210fddef48aff1b]
[Mon Nov 9 00:12:40 2026] ? __pfx_nv_drm_event_callback+0x10/0x10 [nvidia_drm 2e615ec2f99f79ae71a6a4d1e210fddef48aff1b]
[Mon Nov 9 00:12:40 2026] nvKmsKapiHandleEventQueueChange+0xf4/0x150 [nvidia_modeset 3fd61761179da386f97f306a38b0e56e9d555b16]
[Mon Nov 9 00:12:40 2026] ? srso_return_thunk+0x5/0x5f
[Mon Nov 9 00:12:40 2026] _main_loop+0x10e/0x160 [nvidia_modeset 3fd61761179da386f97f306a38b0e56e9d555b16]
[Mon Nov 9 00:12:40 2026] ? srso_return_thunk+0x5/0x5f
[Mon Nov 9 00:12:40 2026] ? __pfx__main_loop+0x10/0x10 [nvidia_modeset 3fd61761179da386f97f306a38b0e56e9d555b16]
[Mon Nov 9 00:12:40 2026] kthread+0xe4/0x120
[Mon Nov 9 00:12:40 2026] ? __pfx_kthread+0x10/0x10
[Mon Nov 9 00:12:40 2026] ret_from_fork+0x292/0x330
[Mon Nov 9 00:12:40 2026] ? __pfx_kthread+0x10/0x10
[Mon Nov 9 00:12:40 2026] ret_from_fork_asm+0x1a/0x30
[Mon Nov 9 00:12:40 2026]
[Mon Nov 9 00:12:40 2026] —[ end trace 0000000000000000 ]—
[Mon Nov 9 07:17:28 2026] [drm:nv_drm_atomic_commit [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00002d00] Flip event timeout on head 0
[Mon Nov 9 07:17:31 2026] [drm:nv_drm_atomic_commit [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00002d00] Flip event timeout on head 1
[Mon Nov 9 07:17:31 2026] ------------[ cut here ]------------
I seem to have a similar issue:
[90357.959824] [drm:nv_drm_atomic_commit [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00002100] Flip event timeout on head 0
[90361.031586] [drm:nv_drm_atomic_commit [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00002100] Flip event timeout on head 1
[90361.031881] ------------[ cut here ]------------
[90361.031892] WARNING: CPU: 2 PID: 1334 at /var/lib/dkms/nvidia/620.70.06/build/kernel-open/nvidia-drm/nvidia-drm-crtc.h:371 __nv_drm_handle_flip_event+0x1ca/0x1e0 [nvidia_drm]
[90361.031934] Modules linked in: cpuid macvtap macvlan vhost_net vhost vhost_iotlb tap tun nfsv3 nfs_acl ext4 mbcache jbd2 ntfs3 exfat snd_seq_dummy snd_hrtimer snd_seq xt_CHECKSUM xt_MASQUERADE xt_conntrack ip6t_REJECT ipt_REJECT rpcsec_gss_krb5 auth_rpcgss xt_tcpudp nft_compat nf_nat_tftp nfsv4 nf_conntrack_tftp bridge dns_resolver stp llc nfs lockd grace netfs nf_conntrack_netbios_ns nf_conntrack_broadcast nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 nf_tables nvidia_drm(OE) nvidia_modeset(OE) qrtr rfcomm cmac algif_hash algif_skcipher overlay af_alg drm_display_helper bnep cec rc_core drm_ttm_helper ttm drm_kms_helper video nvidia_uvm(OE) intel_rapl_msr intel_rapl_common intel_uncore_frequency intel_uncore_frequency_common sunrpc binfmt_misc nvidia(OE) skx_edac skx_edac_common nfit libnvdimm x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm mei_wdt snd_soc_avs irqbypass
[90361.032112] crct10dif_pclmul sha512_ssse3 snd_soc_hda_codec ghash_generic sha256_ssse3 ghash_clmulni_intel sha1_ssse3 gcm crypto_null snd_hda_ext_core rapl snd_soc_core intel_cstate snd_hda_codec_realtek btusb snd_hda_codec_generic snd_compress btrtl btintel snd_pcm_dmaengine snd_hda_codec_hdmi aesni_intel snd_hda_scodec_component btbcm btmtk gf128mul intel_uncore crypto_simd snd_usb_audio snd_hda_intel bluetooth pcspkr cryptd snd_intel_dspcfg hp_wmi snd_intel_sdw_acpi snd_usbmidi_lib sparse_keymap platform_profile snd_hda_codec intel_wmi_thunderbolt snd_rawmidi snd_hda_core snd_seq_device nls_ascii wmi_bmof mc snd_hwdep nls_cp437 vfat ecdh_generic fat snd_pcm rfkill crc16 mei_me snd_timer mei snd soundcore joydev sg msr evdev i2c_dev parport_pc ppdev lp drm parport dm_mod efi_pstore configfs nfnetlink lz4 zram lz4hc_compress lz4_compress ip_tables x_tables autofs4 btrfs blake2b_generic raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c crc32c_generic raid1 raid0 md_mod
[90361.032313] sr_mod hid_generic cdrom sd_mod uas usbhid usb_storage hid ahci libahci xhci_pci nvme libata iTCO_wdt xhci_hcd intel_pmc_bxt nvme_core iTCO_vendor_support igb watchdog psmouse crc32_pclmul usbcore crc32c_intel nvme_auth serio_raw e1000e scsi_mod i2c_algo_bit i2c_i801 ioatdma i2c_smbus usb_common scsi_common wmi dca button efivarfs
[90361.032410] CPU: 2 UID: 0 PID: 1334 Comm: nvidia-modeset/ Tainted: G W OE 6.12.115+deb13.1-amd64 #1 Debian 6.12.115-2
[90361.032426] Tainted: [W]=WARN, [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
[90361.032431] Hardware name: HP HP Z4 G4 Workstation/81C5, BIOS P61 v02.97 06/17/2025
[90361.032437] RIP: 0010:__nv_drm_handle_flip_event+0x1ca/0x1e0 [nvidia_drm]
[90361.032469] Code: 75 00 48 85 f6 74 a7 4c 89 e7 e8 b1 f3 ff ff 80 3d f2 99 00 00 00 74 96 4c 89 e7 e8 30 0a eb fd eb 8c 48 89 df e8 86 09 39 d8 <0f> 0b e9 a8 fe ff ff 48 89 df e8 17 0a eb fd e9 3e ff ff ff 66 90
[90361.032479] RSP: 0018:ffffcf8acad87dd8 EFLAGS: 00010282
[90361.032488] RAX: ffff8bc61fac9818 RBX: ffff8bc61fac9848 RCX: ffff8bc62a7afae0
[90361.032495] RDX: 0000000000000001 RSI: 0000000000000001 RDI: ffff8bc61fac9848
[90361.032501] RBP: ffff8bc61fac9818 R08: 0000000000000000 R09: 0000000000000000
[90361.032507] R10: ffff8bc642ad7c40 R11: 6e6168633d4e4f49 R12: ffff8bc61fac9800
[90361.032513] R13: ffffffffc2e4a818 R14: ffff8bc6499f0000 R15: ffff8bc62a7af800
[90361.032519] FS: 0000000000000000(0000) GS:ffff8bcb67f00000(0000) knlGS:0000000000000000
[90361.032526] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[90361.032532] CR2: 00001c2c0009c000 CR3: 000000043fe22003 CR4: 00000000003726f0
[90361.032539] Call Trace:
[90361.032546]
[90361.032559] nv_drm_event_callback+0xb7/0xd0 [nvidia_drm]
[90361.032587] nvKmsKapiHandleEventQueueChange+0xb7/0x110 [nvidia_modeset]
[90361.032760] ? kfree+0x2ea/0x350
[90361.032775] _main_loop+0x8f/0x150 [nvidia_modeset]
[90361.032921] ? __pfx__main_loop+0x10/0x10 [nvidia_modeset]
[90361.033064] kthread+0xcf/0x100
[90361.033075] ? __pfx_kthread+0x10/0x10
[90361.033085] ret_from_fork+0x31/0x50
[90361.033098] ? __pfx_kthread+0x10/0x10
[90361.033106] ret_from_fork_asm+0x1a/0x30
[90361.033123]
[90361.033127] —[ end trace 0000000000000000 ]—
Linux driver 620.70.06
Card : NVIDIA Corporation GA106 [GeForce RTX 3050 OEM]
uname -a
Linux usb3 6.18.12+deb14-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.18.12-1
After the upgrade I experienced flickering when I played online games with Firefox.
I had to disable the hardware acceleration in Firefox to get rid of it :(
Indiana Jones is still broken on this driver. Details are below with the help of Claude. Proton log and Nvidia crash log are attached.
Indiana Jones and the Great Circle (Steam 2677660): crash on first in-game frames with Proton Wayland on KWin; works in Gamescope. RTX 4090
Error (game log):
FATAL ERROR: vkAcquireNextImageKHR failed with error (VK_NOT_READY) followed by FORCED CRASH (game’s own abort handler)
Crashes: PROTON_ENABLE_WAYLAND=1 %command% (also with DISPLAY="" %command%). Loads vatican_intro, reaches GAME_STATE_INGAME, dies ~0.5s later (~10s after launch).
Works: XWayland (no PROTON_ENABLE_WAYLAND), but no HDR. Gamescope works with Wayland + HDR.
System:
- GPU: RTX 4090, driver 620.70.06 (open kernel module)
- CachyOS, kernel 7.3.3-1-cachyos
- KDE Plasma, KWin Wayland (HDR infoframe active)
- Proton-CachyOS cachyos-11.0-20261015-slr (winewayland.drv)
- Native Vulkan game + Streamline (DLSS/DLSS-G/Reflex), 1920x1080 fullscreen
Findings:
- No Xid/NVRM errors in dmesg, no KWin messages at crash time, so not a GPU/driver fault
- Game treats VK_NOT_READY from vkAcquireNextImageKHR as fatal
- Only fails on the direct Wine Wayland WSI to KWin path; Gamescope’s WSI path is fine
Question: Can the NVIDIA Vulkan driver return VK_NOT_READY from vkAcquireNextImageKHR on a Wayland surface with a zero timeout (e.g. before the first buffer is committed)? Any known issues on KWin Wayland?
nvidia-bug-report.log.gz (888.5 KB)
Has anyone encountered the Displayport sleep bug with this version that was found in the last driver that many reported?
I can confirm that the regression still exist.
It is refresh rate related for me. Works fine up to 144 Hz, not at 240 Hz, using a 3440x1440 UWQHD monitor.
Nvidia-powerd is still quite problematic - whether it is Balanced or Performance mode, it loves to give too much power to the CPU and overutilize it, and I feel like it underutilizes GPU under some GPU-load threshold as well. For instance at 90% load you would still gaing performance for allocating less power to the CPU and more to the GPU, but instead the CPU works its hardest and sits withing 90 degree range with fresh and healthy cooling, and the GPU would be at 60W. Throttling with cpupower brings temperatures back to normal and GPU power to full utilizations. Games also see an improvement in performance, so it is not some CPU bottleneck scenario
Another issue is that it effectively makes Balanced and Performance laptop modes of my Lenovo Legion only differ in cooling levels, the GPU still caps out at around 80W in performance mode, and CPU works as hard as it can in Balanced mode.
At least since 615 it does not cause stutters in games at all, so this is great. I previously reported that you could spot some stutters on the vsynctester website, but as of today I don’t see them anymore.
Using the latest 620 driver with cachyOS kernel 7.3.3 and when trying to launch steam entire system refreezes needing forced shutdown. Reverted back to nvidia version 615 and steam works fine.
Had the exact same problem yesterday. But instead of crashing during Steam launch, it happened after I played a game for 1+ hours, then tried to close Steam.
Regression in 620.70.06: full screen video causes a black screen, then a heavy red tint on one monitor (worked fine on 615)
System
- Driver: 620.70.06 (open kernel module). 615.71.10 does not have this problem.
- OS: Debian kernel 7.2.15
- Compositor: Hyprland 0.57.0 built from branch v0.57.0 at commit 0aa6745f705f2c4b0a5004eb019ef76e5bbdb426 clean (VERSION: bump to 0.57.0), Wayland
- Monitors:
- 43" HDR-capable (affected)
- 27" rotated to portrait (not affected)
- 28" rotated to portrait (not affected)
Problem
When I make a video fullscreen on the 43" monitor, the screen goes black for about a second. The picture then comes back with a heavy red tint over everything. When I leave fullscreen, the screen goes black again for a few seconds and then returns with normal colors.
Fullscreen video on the two rotated monitors works normally.
On 615 the same setup worked perfectly. Entering and leaving fullscreen was practically instant, with no color problems. The only change was upgrading the driver to 620.
Cause narrowed down
The problem is triggered by Hyprland’s automatic HDR switching for fullscreen content (render.cm_auto_hdr). Setting cm_auto_hdr = 0 fixes it completely: no black screen, no tint.
These had no effect on the problem:
render.direct_scanout = 0misc.vrr = 0quirks.skip_non_kms_dmabuf_formats = 1
So the problem seems to be in how 620 handles the switch into and out of HDR output (the HDR metadata or the color format sent to the monitor), not in direct scanout or VRR.
Steps to reproduce
- Hyprland on Wayland with
cm_auto_hdrenabled (the default), driver 620.70.06. - Play a video and make it fullscreen on an HDR-capable monitor.
- The screen blanks for about a second and then shows a strong red tint.
- Leave fullscreen. The screen blanks for a few seconds and then colors are normal again.
I’m experiencing the same display failure with NVIDIA driver 620.70.06 on Fedora 44 with KDE Plasma.
The failure occurs when switching an external monitor’s input source (mine is a MacBook) back to the Fedora laptop. The NVIDIA DRM driver reports flip event timeouts on both display heads, followed by repeated nv_flip == NULL kernel warnings.
Depending on the circumstances, the desktop either:
freezes for several seconds, recovers, and leaves the second monitor disabled
OR
both external displays remain black following the transition
Importantly, starting a Fedora session with both external monitors connected and active works normally. The failure is specifically associated with switching display inputs while the Fedora session is already running.
The laptop’s internal display is disabled.
Two external monitors are used:
-
Monitor 1: Connected to via HDMI and continuously displaying the Fedora laptop.
-
Monitor 2: Has two independent input sources: one connected to the Fedora laptop via USB-C (DP) and one connected to a MacBook via HDMI.
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: [ 2555][904.700812] (dw_udev_watch ) udev event detected
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: [ 2555][904.700966] Udev_Event_Detail
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_subsystem: drm
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_action: change
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_connector: (null)
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_devname: /dev/dri/card0
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_devmode: (null)
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_hotplug: 1
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_major: 226
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: prop_minor: 0
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: sysname: card0
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: syspath: /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0/drm/card0
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: [ 2555][904.701027] (dw_pause_retaking_if_suspended) Allowing time for events to coalesce: Sleeping for 100 milliseconds
Nov 10 13:25:33 ultramarine org_kde_powerdevil[2331]: [ 2555][904.801143] (dw_udev_drain ) Drained 0 additional queued udev event(s)
Nov 10 13:25:37 ultramarine kernel: [drm:nv_drm_atomic_commit [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Flip event timeout on head 0
Nov 10 13:25:40 ultramarine kernel: [drm:nv_drm_atomic_commit [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Flip event timeout on head 1
Nov 10 13:25:40 ultramarine kernel: ------------[ cut here ]------------
Nov 10 13:25:40 ultramarine kernel: nv_flip == NULL
Nov 10 13:25:40 ultramarine kernel: WARNING: nvidia-drm/nvidia-drm-crtc.h:371 at __nv_drm_handle_flip_event+0x1f9/0x210 [nvidia_drm], CPU#12: nvidia-modeset//1047
Nov 10 13:25:40 ultramarine kernel: CPU: 12 UID: 0 PID: 1047 Comm: nvidia-modeset/ Tainted: G S W OE 7.3.3-1.cachyos1.fc44.x86_64 #1 PREEMPT(full) 931b624e1e26b64e15a2968ff5f25b06cf18b89d
Nov 10 13:25:40 ultramarine kernel: Tainted: [S]=CPU_OUT_OF_SPEC, [W]=WARN, [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
Nov 10 13:25:40 ultramarine kernel: Hardware name: LENOVO 82TD/LNVNB161216, BIOS K1CN48WW 07/10/2024
Nov 10 13:25:40 ultramarine kernel: RIP: 0010:__nv_drm_handle_flip_event+0x1f9/0x210 [nvidia_drm]
Nov 10 13:25:40 ultramarine kernel: Code: 6c 24 20 4c 8b 64 24 28 4c 8b 74 24 38 48 83 c4 48 e9 ab 66 00 00 4c 8b 6c 24 30 4c 8b 7c 24 40 eb af 4c 89 e7 e8 37 14 5e ef <0f> 0b eb 9b 48 8d bd b0 00 00 00 e8 87 1
Nov 10 13:25:40 ultramarine kernel: RSP: 0018:ffffc8ed43353dc8 EFLAGS: 00010282
Nov 10 13:25:40 ultramarine kernel: RAX: ffff88e458ca3818 RBX: ffff88e458ca3818 RCX: ffff88e45a22e2c0
Nov 10 13:25:40 ultramarine kernel: RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff88e458ca3840
Nov 10 13:25:40 ultramarine kernel: RBP: ffff88e458ca3800 R08: 0000000000000000 R09: ffffffffc28cf318
Nov 10 13:25:40 ultramarine kernel: R10: fffff649476c5180 R11: ffff88e440042700 R12: ffff88e458ca3840
Nov 10 13:25:40 ultramarine kernel: R13: ffff88e4607dd4c0 R14: ffff88e45a22e000 R15: ffffffffc22a15c0
Nov 10 13:25:40 ultramarine kernel: FS: 0000000000000000(0000) GS:ffff88f3fa216000(0000) knlGS:0000000000000000
Nov 10 13:25:40 ultramarine kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Nov 10 13:25:40 ultramarine kernel: CR2: 00001e14016bb000 CR3: 0000000cc1422004 CR4: 0000000000f72ef0
Nov 10 13:25:40 ultramarine kernel: PKRU: 55555554
Nov 10 13:25:40 ultramarine kernel: Call Trace:
Nov 10 13:25:40 ultramarine kernel: <TASK>
Nov 10 13:25:40 ultramarine kernel: ? __pfx__main_loop+0x10/0x10 [nvidia_modeset 3529bee662f5c7aeefd64613572afbfef22ab594]
Nov 10 13:25:40 ultramarine kernel: nv_drm_event_callback+0xaa/0xb0 [nvidia_drm 44bfb97dd735dba7f534e950d80012f19402e976]
Nov 10 13:25:40 ultramarine kernel: nvKmsKapiHandleEventQueueChange+0xba/0x110 [nvidia_modeset 3529bee662f5c7aeefd64613572afbfef22ab594]
Nov 10 13:25:40 ultramarine kernel: _main_loop+0x111/0x170 [nvidia_modeset 3529bee662f5c7aeefd64613572afbfef22ab594]
Nov 10 13:25:40 ultramarine kernel: kthread+0xe6/0x120
Nov 10 13:25:40 ultramarine kernel: ? __pfx_kthread+0x10/0x10
Nov 10 13:25:40 ultramarine kernel: ret_from_fork+0x2a9/0x340
Nov 10 13:25:40 ultramarine kernel: ? __pfx_kthread+0x10/0x10
Nov 10 13:25:40 ultramarine kernel: ret_from_fork_asm+0x1a/0x30
Nov 10 13:25:40 ultramarine kernel: </TASK>
Nov 10 13:25:40 ultramarine kernel: ---[ end trace 0000000000000000 ]---
Nov 10 13:25:40 ultramarine kernel: ------------[ cut here ]------------
Nov 10 13:25:40 ultramarine kernel: nv_flip == NULL
Nov 10 13:25:40 ultramarine kernel: WARNING: nvidia-drm/nvidia-drm-crtc.h:371 at __nv_drm_handle_flip_event+0x1f9/0x210 [nvidia_drm], CPU#12: nvidia-modeset//1047
Nov 10 13:25:40 ultramarine kernel: CPU: 12 UID: 0 PID: 1047 Comm: nvidia-modeset/ Tainted: G S W OE 7.3.3-1.cachyos1.fc44.x86_64 #1 PREEMPT(full) 931b624e1e26b64e15a2968ff5f25b06cf18b89d
Nov 10 13:25:40 ultramarine kernel: RIP: 0010:__nv_drm_handle_flip_event+0x1f9/0x210 [nvidia_drm]
Nov 10 13:25:40 ultramarine kernel: RSP: 0018:ffffc8ed43353dc8 EFLAGS: 00010282
Nov 10 13:25:40 ultramarine kernel: RAX: ffff88e458760818 RBX: ffff88e458760818 RCX: ffff88e45a22e2c0
Nov 10 13:25:40 ultramarine kernel: RDX: 0000000000000001 RSI: 0000000000000001 RDI: ffff88e458760840
Nov 10 13:25:40 ultramarine kernel: RBP: ffff88e458760800 R08: 0000000000000000 R09: ffffffffc28cf318
Nov 10 13:25:40 ultramarine kernel: R10: fffff64944049f40 R11: ffff88e440042700 R12: ffff88e458760840
Nov 10 13:25:40 ultramarine kernel: R13: ffff88e4607dd4c0 R14: ffff88e45a22e000 R15: ffffffffc22a15c0
Nov 10 13:25:40 ultramarine kernel: PKRU: 55555554
Nov 10 13:25:40 ultramarine kernel: ---[ end trace 0000000000000000 ]---
Nov 10 13:25:40 ultramarine xdg-desktop-portal-kde[2753]: PipeWire remote error: -116 no global 121 any more
Nov 10 13:25:40 ultramarine plasmashell[2184]: requesting unexisting screen available rect 1
Nov 10 13:25:40 ultramarine plasmashell[2184]: requesting unexisting screen geometry 1
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: [ 2555][912.458410] (dw_stabilized_buses_bs) Required 2 extra 100 millisecond calls to i2c_buses_w_edid_as_bitset()
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: [ 2555][912.458447] (dw_remove_display_by_businfo) Removing connected display on bus 21
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: [ 2555][912.458450] (dw_remove_display_by_businfo) (dw_remove_display_by_businfo) No Display_Ref found for i2c bus: 21
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: After buses added or removed:
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: Number of buses: 18
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: All I2C buses: 6 7 8 9 10 11 12 13 14 15 16 17 19 20 21 22 23 24
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: Buses with edid: 15 22
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: Buses with x37 active: 22
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: After display refs added or marked disconnected:
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: Display_Ref[1:bus /dev/i2c-15 @0x56226fac0cc0]
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: Display_Ref[Disconnected: 2:bus /dev/i2c-21 @0x56226faef370]
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: Display_Ref[3:bus /dev/i2c-22 @0x56226facb4b0]
Nov 10 13:25:41 ultramarine org_kde_powerdevil[2331]: [ 2555][912.458542] (dw_udev_drain) Drained 2 additional queued udev event(s)
Nov 10 13:25:44 ultramarine kernel: nvidia-modeset: WARNING: GPU:0: Lost display notification (0:0x00000000); continuing.
S3 suspend/resume: deferred fbcon takeover vs nvidia-modeset rwsem deadlock — is this addressed in 620.70.06? (RTX 5070, GNOME Wayland)
Environment: RTX 5070 (GB205) desktop on Arrow Lake-S (ASUS ROG MAXIMUS Z890 HERO, Core Ultra 7 265K), Ubuntu 26.04.1, kernel 7.0.0-43-generic, GNOME Wayland, open modules 615.71.09, S3 (deep), NVreg_PreserveVideoMemoryAllocations=1 with the distribution’s nvidia-suspend/nvidia-resume systemd services.
I’d like to know whether the following deadlock class is fixed or known-fixed in 620.70.06, since the alternative is upgrading a production machine blind:
During suspend/resume, the deferred fbcon takeover work races the nvidia-modeset rwsem and the console lock, producing an unrecoverable AB-BA (recovery = SysRq+B). I’ve reproduced it 6× on 615.71.09 across three configurations (kernel suspend notifiers on/off × user-session freezing on/off):
codeblock
kworker (either resume or entry phase):
Workqueue: events fbcon_register_existing_fbs
down_read → nvkms_ioctl_from_kapi → GetDynamicDisplayInfo [nvidia_modeset]
(via nv_drm_connector_detect / fbcon_init)
...blocked on the nvidia-modeset rwsem, WHILE HOLDING console_lock
systemd-sleep, one of:
- notifiers=1: nv_pm_notifier → nvidia_resume → rm_power_management (holds the rwsem write; blocked on a semaphore the kworker last held)
- notifiers=0: enter_state → pm_restore_console → vt_move_to_console → console_lock (blocked on the console lock the kworker holds)
Observations that may help triage:
- In one entry-side occurrence, the FBSR save (
nvidia-suspend.service→/proc/driver/nvidia/suspend) completed successfully at 17:53:59, and the takeover kworker blocked on the rwsem from 18:10:19 — with no writer visible in any hung-task report or the SysRq show-state dump. The lock appears to be left held by the suspend-side path; this matches issue #1281’s “PM lock permanently held after S3”. On cycles where the lock is free, the same kworker runs the same path instantly and resume is completely clean. - With
NVreg_UseKernelSuspendNotifiers=0(preserve AUTO → off), the deadlock disappears, but every resume loses VRAM: 5×/12× Xid 13 at the resume instant, gnome-shell channel faults, “Flip event timeout on head 0”, Xid 31/154. So there is no working configuration — the deadlock-free one breaks the desktop instead. SYSTEMD_SLEEP_FREEZE_USER_SESSIONS=true(freezing user sessions) does not prevent the deadlock, even though it removes live userspace GPU clients from the window.- Entries frequently show a silent stall between “PM: suspend entry” and “Freezing user space processes” — usually 2–3.5 min, once 15 min 19 s.
Related reports: #1281 (PM lock left held after S3, RTX 5070), #1291 (nvidia-sleep.sh path), #1329 (client in nv_drm_mmap holding the rwsem during BAR2 teardown); Ubuntu LP #2166914 and LP #2171636 (identical fbcon/console_lock ↔ rwsem cycle on other GPUs/drivers).
Two questions: (a) has the rwsem/console interaction around the FBSR save/restore window been reworked in 620.70.06? (b) if not, is there a supported way to keep the deferred fbcon takeover from doing connector detection inside the driver’s PM window on Wayland systems — where the Xorg-motivated chvt 63 / no-freeze conventions in nvidia-sleep.sh don’t apply?
On latest cachyos with 620 drivers and KDE plasma, cyberpunk 2077 hangs when trying to do a benchmark. Machine is AMD 5900x and 3080ti GPU.